# Silent Payments: Light Client Protocol

**URL:** <https://delvingbitcoin.org/t/silent-payments-light-client-protocol/891>\
**Category:** wg-silent-payments\
**Tags:** wallet\
**Created:** [May 21, 2024, 9:15am UTC](https://delvingbitcoin.org/t/silent-payments-light-client-protocol/891 "2024-05-21T09:15:50Z")\
**Posts on this page:** 1\
**Showing post:** 9

<div class="post-metadata">

**Author:** ![setavenger](https://delvingbitcoin.org/user_avatar/delvingbitcoin.org/setavenger/32/573_2.png) [@setavenger](https://delvingbitcoin.org/u/setavenger)\
**Post date:** [May 23, 2024, 10:04am UTC](https://delvingbitcoin.org/t/silent-payments-light-client-protocol/891/9 "2024-05-23T10:04:41Z")

</div>

> [@cygnet3](#):
>
> When testing out Blindbit oracle as a backend, we just convert the utxo array into a map and then loop over the values. That has the same structure as the json you posted. It could be simplified by removing `txid` from the output struct.

Is there any noticeable overhead created by converting on the client side? I think we should benchmark these approaches to see how they compare. Then we can better say whether grouping should be a hard requirement on the indexing side.

> [@cygnet3](#):
>
> If not using blindbit, we use a BIP158 client and request the full block, so that has the complete transaction structure.

How does that approach work in general? Something like this?

1. Fetch tweaks + Filters
2. Create ScriptPubKeys (with mapping)
3. Find match
4. Download entire block
5. Scan transactions and find outputs

Where step 5 is optimised due to mapping and grouping. Did I get this right?

You mentioned that you’ve tried BlindBit Oracle as well, are there any noticeable differences in required bandwidth? With BlindBit the goal was to avoid downloading entire blocks to reduce bandwidth, but I never made an actual comparison.

---

_[View the full topic](https://delvingbitcoin.org/t/silent-payments-light-client-protocol/891)._
